JP2020523663A - Information retrieval method, device and system - Google Patents

Abstract

The present invention provides an information search method, device and system. The information search method receives a search word transmitted from a user via a client (SlOl), and identifies the user's search intention based on the characteristic information of the search word and/or the search context characteristic information (SlOl). SlO2), using a search policy corresponding to the identified search intention and performing a search based on the above search word, and obtaining an information search result related to the search word (SlO3), and the information search result To the client and display the information retrieval result on the client (SlO4).

Description

<Cross reference to related application>
This patent application claims the priority of a Chinese patent application filed on July 20, 2017, whose application number is 2017105968799.4 and whose invention title is “information retrieval method, device and system”, The entire contents of the application are incorporated into this application by reference.

The present invention relates to the field of information retrieval technology, and more particularly, to an information retrieval method, device and system.

BACKGROUND OF THE INVENTION With the rapid development of Internet of Things technology, offline attractions, merchants, etc. are now being discovered and understood by many people via the Internet online. A user searches for offline attractions, merchants, etc. through an O2O (Online To Offline: Internet concept of things that combines offline business opportunities and the Internet) search engine using terminals such as computers and mobile phones, and also attraction information. Understand information such as, merchant promotion activities, consumer user comments, or perform services such as online ticket booking, ordering, and payment.

Currently, the related art O2O search engine mainly searches based on local information, and when a user searches for a national chain hotel, for example, only local search results can be obtained, that is, the current location. It is possible to provide the user with only the relevant search results of the existing city.

However, in practical applications, users often have many off-site search needs, such as querying destination city attraction information, hotel information and ticket booking, hotel reservation before travelling. Although it is necessary to provide services such as reservations, conventional O2O search engines generally cannot meet user needs. For example, when a user searches for "Huashan" in Shanghai, the results of "Huashan Hospital" are displayed. As a result, if the user wants to search for the "Huashan" attraction, he has to manually switch between cities, which is cumbersome and inefficient.

Therefore, there is currently an urgent need for a technical solution that can solve the user's off-site search needs in order to increase the efficiency of the user's off-site search and improve the user experience.

In view of this, it is an object of an embodiment of the present invention to provide an information search method, device and system.

According to a first aspect, an information search method for a server according to an embodiment of the present invention includes receiving a search word transmitted from a user via a client, and using the search word feature information and/or the search context feature. Identifying the user's search intent based on the information, and using the search policy corresponding to the identified search intent and searching based on the search term to obtain information search results related to the search term. Here, the information search result includes a local search result and/or an off-site search result, and the information search result is transmitted to the client and the client displays the information search result.

Preferably, identifying the search intention of the user based on the characteristic information of the search word and/or the search context characteristic information includes acquiring the characteristic information of the search word and/or the search context characteristic information, and Determining the off-site search preference level of the user by a pre-trained off-site search preference level determination model based on the search word feature information and/or the search context feature information; Determining the search intention of the user based on the set preference threshold value condition.

Preferably, the off-site search preference level determination model uses the feature variable of the search word and/or the search context feature variable as an independent variable, the off-site search preference feature as a dependent variable, and the off-site search preference level determination model It is obtained by training by building, obtaining training samples generated based on a plurality of search click history records, and training the off-site search preference model with the training supplements.

Preferably, constructing the off-site search preference degree determination model includes constructing the off-site search preference degree determination model using a logical regression model.

Preferably, constructing the off-site search preference degree determination model using a logical regression model specifically includes constructing the off-site search preference degree determination model by the following Expression 1.

Preferably, before identifying the user's search intent based on the search term feature information and/or the search context feature information, the method further comprises presetting the search terms in local and offsite entity word libraries. To determine whether or not the search word has a possibility of off-site search, and if there is a possibility of off-site search, the characteristic information of the search word and/or the search context characteristic information are Performing a step of identifying the user's search intent.

Preferably, using the search policy corresponding to the search intent and performing a search based on the search term, and obtaining information search results related to the search term, the search intent is determined by the user to be local and off-site. In the case of having the search intention at the same time, it includes performing a local search and an off-site search based on the search word, and obtaining the information search result including a plurality of local search results and a plurality of off-site search results.

Preferably, before sending the information search results to the client, the method further merges the plurality of local search results into a local search result display area, and the plurality of offsite search results is displayed in an offsite search result display area. And deciding the display order of the local search result display area and the off-site search result display area based on the user's off-site search preference and/or the quality of local and off-site search results. Generating the information search result including the local search result display area and the off-site search result display area based on the display order.

Preferably, the local search result display area is a foldable display area, and after merging the plurality of local search results into the local search result display area, the method further includes calculating a predicted click rate for each of the local search results. And determining the order of the local search results in the local search result display area, and determining the number of the local search results displayed in the folded state of the local search result display area.

Preferably, the off-site search result display area is a foldable display area, and after merging the plurality of off-site search result display areas with the off-site search result display area, the method further includes predicting each of the off-site search results. Establishing the order of each of the off-site search results in the off-site search result display area based on the click rate, and determining the number of the off-site search results displayed in the folded state of the off-site search result display area. Can be included.

Preferably, the information search method further includes acquiring a user's click operation log for the information search result, and iteratively optimizing the off-site search preference degree determination model based on the click operation log. be able to.

In a second aspect, an information retrieval method for a client according to an embodiment of the present invention comprises intercepting a search term input by a user, sending the search term to a server, and providing feedback from the server. Received information retrieval results, wherein the information retrieval results include a retrieval policy corresponding to the user's retrieval intention identified by the server based on the retrieval word characteristic information and/or the retrieval context characteristic information. It has been obtained by using and searching, wherein the information search result includes a local search result and/or an off-site search result, and displaying the information search result.

Preferably, the information search result may include a local search result display area, an off-site search result display area, and a display order. Here, the local search result display area may include one or a plurality of local search results, and the off-site search result display area may include one or a plurality of off-site search results. .. Correspondingly, displaying the information search result may include displaying the local search result display area and the off-site search result display area on the same screen in the display order.

In a third aspect, an information search device for a server according to an exemplary embodiment of the present invention includes a search word receiving module for receiving a search word transmitted from a user through a client, feature information of the search word, and And/or using a search intention identifying module for identifying the search intention of the user based on the search context characteristic information, and a search policy corresponding to the search intention and performing a search based on the search word to obtain the search word Used to obtain relevant information search results, wherein the information search results include a policy search module including a local search result and/or an off-site search result, and the information search result is transmitted to the client, And a search result transmission module for displaying the information search result on a client.

Preferably, the search intention identifying module specifically acquires the feature information and/or the search context feature information of the search word, and based on the feature information of the search word and/or the search context feature information, Determining the off-site search preference degree of the user by a pre-trained off-site search preference degree determination model, and determining the user's search intention based on the off-site search preference degree and a preset preference degree threshold condition. Used to determine.

Preferably, the off-site search preference level determination model uses the feature variable of the search word and/or the search context feature variable as an independent variable, the off-site search preference feature as a dependent variable, and the off-site search preference level determination model It is obtained by training by building, obtaining training samples generated based on a plurality of search click history records, and training the off-site search preference model with the training supplements.

Preferably, the off-site search preference degree determination model is constructed using a logical regression model.

Preferably, the off-site search preference determination model is specifically constructed by the following Expression 1.

Preferably, the information search device further matches the search term with preset local and off-site entity word libraries to determine whether the search term has a possibility of off-site search. If there is a possibility of off-site search, an entity word matching module is included for performing the step of identifying the user's search intention based on the search word characteristic information and/or the search context characteristic information.

Preferably, the policy search module specifically performs a local search and an offsite search based on the search term when the search intent is that the user has a local search intent and an offsite search intent at the same time. , Used to obtain the information search results including a plurality of local search results and a plurality of off-site search results.

Preferably, the information search device further merges the plurality of local search results into a local search result display area, and a result merge module for merging the plurality of off-site search results into an off-site search result display area, A display order determination module for determining a display order of the local search result display area and the offsite search result display area based on a user's off-site search preference and/or quality of local and off-site search results; And a search result generation module for generating the information search result including the local search result display area and the off-site search result display area based on a display order.

Preferably, the local search result display area is a foldable display area. In this case, the information search device further determines the order of the local search results in the local search result display area based on the predicted click rate of each of the local search results, and in the folded state of the local search result display area. A local order confirmation module for determining the number of displayed local search results may be provided.

Preferably, the off-site search result display area is a foldable display area. In this case, the information search device further determines the order of the off-site search results in the off-site search result display area based on the predicted click rate of each of the off-site search result, and the off-site search result display area. An off-site order confirmation module for determining the number of the off-site search results displayed in the collapsed state can be provided.

Preferably, the information search device further optimizes an operation log acquisition module for acquiring a click operation log of a user for the information search result, and the off-site search preference determination model based on the click operation log. And an iterative optimization module for

In a fourth aspect, an information search device for a client according to an exemplary embodiment of the present invention includes an input monitoring module for monitoring a search word input by a user, and a search word for transmitting the search word to a server. The sending module is used to receive the information search results fed back from the server, the information search results being searched by the server based on the search word feature information and/or the search context feature information. The information retrieval result is obtained by identifying the intention and then performing a search using a search policy corresponding to the identified search intention, and the information retrieval result includes a local retrieval result and/or an off-site retrieval result. And a search result display module for displaying the information search result.

Preferably, the information search result may include a local search result display area, an off-site search result display area, and a display order. Here, the local search result display area may include one or a plurality of local search results, and the off-site search result display area may include one or a plurality of off-site search results. .. Correspondingly, the search result display module may be specifically used to display the local search result display area and the off-site search result display area on the same screen in the display order.

In a fifth aspect, an information retrieval system according to an embodiment of the present invention comprises a server and a client, the server includes the device described in the third aspect, and the client is the device described in the fourth aspect. including.

In a sixth aspect, a computer device according to an embodiment of the present application comprises a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the first program is executed when the processor executes the computer program. Performing the steps of the method described in the aspects.

In a seventh aspect, a computer device according to an embodiment of the present invention comprises a memory, a processor, and a computer program stored in the memory and executable by the processor, and the second computer program is executed when the processor executes the computer program. Performing the steps of the method described in the aspects.

A computer-readable storage medium according to an embodiment of the present invention according to an eighth aspect stores a computer program for performing the steps of the method according to the first aspect when the computer program is executed by a processor.

A computer readable storage medium according to an embodiment of the present invention according to a ninth aspect stores a computer program for performing the steps of the method according to the second aspect when the computer program is executed by a processor.

An information search method for a server according to an embodiment of the present invention is to receive a search word transmitted from a user through a client, and based on the search word characteristic information and/or the search context characteristic information, the user's Identifying the search intent, using a search policy corresponding to the search intent and performing a search based on the search term, and obtaining an information search result related to the search term; And transmitting the information retrieval result to the client. The present invention identifies a user's search intention and searches according to a corresponding search policy, for example, performing a local search and an off-site search at the same time or performing only an off-site search to obtain a search result that matches the user's intention. This allows off-site search results to be obtained without switching the city where the user is currently located when the user has an off-site search intent, thereby improving search efficiency and effectively improving the user experience. Can be made.

In order to make the above objects, features and advantages of the present invention clearer and easier to understand, the following detailed description will be given with particular reference to the preferred embodiments.

In order to more clearly describe the technical solutions in the embodiments of the present invention, the drawings necessary for the embodiments will be briefly described below. It should be noted that the following drawings are only for showing some embodiments of the present invention, and should not be considered as limiting the scope, and those skilled in the art will need creative efforts. Instead, other related drawings can be obtained based on these drawings.
5 is a flowchart of an information retrieval method for a server according to an exemplary embodiment of the present invention. FIG. 6 is a structural diagram of an information retrieval device for a server according to another embodiment of the present invention. It is a module block diagram of the computer apparatus in the further Example of this invention. 5 is a flowchart of an information retrieval method for a client according to an exemplary embodiment of the present invention. FIG. 6 is a structural diagram of an information retrieval device for a client according to another embodiment of the present invention. It is a module block diagram of the computer apparatus in the further Example of this invention. 1 is a structural diagram of an information retrieval system according to an embodiment of the present invention.

Hereinafter, the technical solutions in the embodiments of the present application will be clearly and completely described in combination with the drawings of the embodiments of the present application, but obviously, the described embodiments are only a part of the embodiments of the present application. And not all examples. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

Embodiments of the present invention provide an information retrieval method, device, system, computer device and computer-readable storage medium. An example will be described below.

As shown in FIG. 1, the first embodiment of the present invention provides an information retrieval method. The execution body of the method is a server, and the method includes steps S101-S104, and is specifically shown below.
In step S101, the search word transmitted from the user via the client is received.
In step S102, the user's search intention is identified based on the search word characteristic information and/or the search context characteristic information.
In step S103, a search policy corresponding to the search intention is used and a search is performed based on the search word, and an information search result related to the search word is obtained, where the information search result is a local search result and Includes offsite search results.
In step S104, the information search result is transmitted to the client, and the client is caused to display the information search result.

In the first embodiment of the present invention, a user inputs a search word to a client, the client monitors the input search word, and then sends the search word to a server for searching, and the server searches the search word. After receiving, the search intention is first identified to determine the search intention of the user, and then the search is performed using the corresponding search policy according to the search intention of the user.

In the embodiment of the present invention, the local and off-site are all for the city where the client is currently located, that is, the city or area where the client is currently located is local and the non-local area is off-site. Is regarded as

The search intent is that the user has an offsite search need and the user does not have an offsite search need so as to determine whether to perform an offsite search and display the offsite search results. Can be included. Here, the user having the off-site search needs further means that the user has the local search needs and the off-site search needs at the same time, for example, the user has a strong off-site search intention, and It may be subdivided into two situations, where the user has only offsite search needs, eg, the user has a very strong offsite search intent. The user does not have an offsite search need means that the user has only local search needs, eg, the user does not have an offsite search intent or has a weak offsite search intent. May be considered.

The above search intent is used to explain that the user wants to query local merchants, offsite merchants, or wants to query local merchants or offsite merchants at the same time, and in the real life the user will get more local merchants. Because of the search, when identifying the user's search intent, the user's search intent can be determined by the strength of the user's off-site search intent. Here, the strength of the user's off-site search intention may be measured based on the off-site search preference, and the user's search may be performed according to the off-site search preference and preset preference threshold conditions. Determine your intentions. For example, if the offsite search preference is lower than the first preference threshold, it can be determined that the user has only local search needs, and the offsite search preference is less than the first preference threshold. If large and smaller than the second preference threshold, it can be determined that the user has both local search needs and offsite search needs at the same time, and the offsite search preference is greater than the second preference threshold. If so, it can be determined that the user has only offsite search needs.

The user's search intention may be determined from two aspects, that is, the strength of the off-site search intention that the search term input by the user has and the strength of the user's off-site search intention of the search context.

Therefore, in one example according to the first embodiment of the present invention, first, the characteristic information of the search term and/or the characteristic information of the search context that can reflect the off-site search intention of the search itself is acquired, The user's search intention is identified based on the characteristic information of the search word and/or the characteristic information of the search context.

The characteristic information of the search word may be a click rate of an off-site search result in the history search result of the search word, an off-site search amount of the search word, or the like, for example, a history search record of one search word. When the click rate of the off-site search result is higher than a preset threshold value, the search term itself may be considered to have a high off-site search intention. If the amount of search in the site city is higher than a preset threshold, then the search term itself may also be considered to have a high off-site search intent.

The feature information of the search context may include user feature information, wherein the user feature information is a history search behavior of the user (for example, whether or not the off-site information is searched), a residence city and a recently active city, Abstract location information such as current location information, history click behavior, history purchase behavior, history browsing behavior, etc., such as the number of city changes within a specified period of time, click rate or number of clicks in offsite search results, offsite It may be the number of orders, etc., and the strength of the user's off-site search intention can be determined based on at least one of the above-described user characteristic information, and the user's search intention can be determined.

Further, the feature information of the search context may include the current local search result quality. Here, the local search result quality may be determined based on at least one of the number of local search results, local search result history clicks, or predicted click rate of the local search results. It is possible to determine the strength of the off-site search intention of the user based on the result quality, and further to determine the search intention of the user. For example, after a search word is received, a local search is first performed and the number of local search results is lower than a preset threshold value of local search results or the local search result history click rate is lower than a preset history click rate threshold value. In this case, the current local search result quality may be determined to be poor, and the user may need to perform an offsite search. Including the two situations of having search needs at the same time and the user having only off-site search needs, which can be distinguished by further presetting threshold conditions) Can be determined.

Note that the above-described characteristic information of the search word and the characteristic information of the search context may be separately used as the basis for identifying the user's search intention, or may be jointly used as the basis for identifying the user's search intention. In step S102, the user's search intention can be identified based only on the characteristic information of the search word, and the search word itself input by the user is high in the off-site search intention (for example, from the first search word off-site intention threshold value). If the user has an offsite search need (the user has local and offsite search needs at the same time, and the user has only an offsite search need). It is possible to directly determine that the two conditions are included and can be distinguished by further setting a threshold condition), and the user search based on only the feature information of the search context in S102. The intention can also be searched, and when the preset threshold condition corresponding to the feature information of at least one of the search contexts described above is satisfied, the user needs the offsite search (the user needs the local search and the offsite search. It can be distinguished by further presetting threshold conditions, including the two situations of having needs at the same time and the user having only off-site search needs). For example, in step S102, the search intention of the user can be jointly identified by combining the feature information of the search word and the feature information of the search context, and the weight corresponding to each identification basis can be determined. Can be set, and each search intention can be weighted to determine the final search intention. When identifying a user's search intent, it is easy to understand that the more the identification basis is, the more accurately the acquired search intent will be obtained, and that the user will be provided with search results that further meet the user's needs to improve the user experience. To be done.

In one embodiment according to the present invention, in order to identify a user's search intention by using a logical regression model, an offsite search preference determination model trained based on the logical regression model is first used to off-site the user. The search preference is calculated, and then the user's search intention is determined based on the off-site search preference. Identifying the user's search intent based on the search word feature information and/or the search context feature information includes acquiring the search word feature information and/or the search context feature information, and Based on the feature information and/or the search context feature information, determining the off-site search preference level of the user by a pre-trained off-site search preference level determination model, and setting the off-site search preference level in advance. Determining the search intention of the user based on the preference threshold condition.

Here, the preference degree threshold condition may include a first preference degree threshold value, and the preset preference degree threshold condition is that the off-site search preference degree is smaller than the first preference degree threshold value and the user's search is performed. It is determined that the intention is that the user does not have an off-site search need, and if the off-site search preference is greater than the first preference threshold, the user's search intention indicates that the user has an off-site search need. It may be to determine that In order to more accurately classify, the preference threshold condition may include a first preference threshold and a second preference threshold, and the preset preference threshold condition is that the off-site search preference is first. If the off-site search preference is greater than the first preference threshold and less than the second preference threshold, then the user can be determined to have only local search needs. Can have both local and offsite search needs at the same time, and if the offsite search preference is greater than the second preference threshold, then the user has only offsite search needs. It may be something that can be confirmed.

In this way, the search intention of the user can be determined based on the off-site search preference of the user and the preset preference threshold condition. Here, the off-site search preference determination model is a logical regression model, in which the search variable feature variable and/or the search context feature variable are independent variables, the off-site search preference feature is a dependent variable, and the off-site search preference is Obtained by being trained by constructing a degree-determining model, obtaining training samples generated based on a plurality of search click history records, and training the off-site search preference degree model with the training supplement. ..

In the off-site search preference degree determination model, the feature value corresponding to the off-site search preference feature that is the dependent variable is the off-site search preference degree, and the feature value corresponding to the search word feature variable is the search word feature information. Yes, the feature value corresponding to the search context feature variable is the search context.

In one embodiment according to the present invention, the off-site search preference level determination model is constructed by using the logical regression model. Specifically, the off-site search preference level determination model is constructed by the following Equation 1. is there.

Based on the logical regression model, the off-site search preference determination model is constructed using various identification grounds (feature variables in the above formula), and then the off-site search is trained using a large number of search click history records. The weight of each feature variable in the preference degree determination model can be determined, and can be used to identify the off-site search preference degree of the user when the training is completed. Specifically, based on the identification basis selected at the time of training (that is, at least one of the characteristic information of the search word and the search context information), the corresponding identification basis (matches the characteristic variable in the model). , The off-site search preference level determination model calculates and outputs a corresponding off-site search preference level based on the identification basis and the trained weights, and the off-site search preference level and the preset preference The user's search intention can be determined by comparing the frequency threshold conditions.

When the user's search intention is identified using the logical regression model based on the description of the above embodiment, more identification bases can be combined, the user's search intention can be identified more accurately, and the user needs can be further improved. It is possible to provide the user with satisfying search results and effectively improve the user experience.

In one embodiment according to the present invention, user query keywords are combined with data sources such as merchant information, statistical rules and machine learning algorithm models are used to establish local and offsite entity word libraries, and potential for offsite consumption. Withdraw potential entity words, such as star hotels, famous attractions, famous brand merchants, etc. Prior to the step of identifying the user's search intent based on the search term feature information and/or the search context feature information, the method further comprises preconfiguring the search term with a local and off-site entity word library. Matching is performed to determine whether the search term has a possibility of off-site search, and if there is a possibility of off-site search, based on the characteristic information of the search term and/or the search context characteristic information. , Performing a step of identifying the user's search intent.

The above steps allow for initial training of the search terms, further identification of the user's search intent if there is an offsite search possibility, and identification of the user's search intent if there is no offsite search possibility. Instead, a local search can be performed directly, thereby improving the overall search efficiency.

If it is determined whether the user has the off-site search intention, the search policy corresponding to the search intention can be further used for the search. For example, if the user does not have an offsite search intent, the corresponding search policy only performs local search, and if the user has an offsite search intent, the corresponding search policy turns off not only local search. It may be a site search. Here, the user's off-site search intention is very strong (the off-site search intention may be considered to be very strong if the off-site search preference exceeds a preset second preference threshold). The corresponding search policy may also be considered to perform only off-site search.

In one embodiment according to the present invention, using the search policy corresponding to the search intent and performing a search based on the search term and obtaining information search results related to the search term means that the search intention is a user. Is to have a local search intention and an off-site search intention at the same time, a local search and an off-site search are performed based on the search word, and an information search result including a plurality of local search results and a plurality of off-site search results is obtained. Including getting.

Considering that the information search result includes not only the local search result but also the off-site search result, mixing the local search result and the off-site search result inevitably confuses the user and complicates the sorting algorithm. Will increase. Therefore, one embodiment according to the present invention adopts the technical solution of displaying local search results and off-site search results in blocks. Specifically, prior to the step of sending the information search results to the client, the method further merges the plurality of local search results into a local search result display area and off-sites the plurality of off-site search results. Merging with the search result display area, and determining the display order of the local search result display area and the offsite search result display area based on the user's off-site search preference and/or quality of the off-site search result. And generating an information search result including the local search result display area and the off-site search result display area based on the display order.

Here, the display order of the local search result display area and the off-site search result display area may be determined based on a user's off-site search preference and/or local and off-site search result quality. Here, the local and off-site search result quality includes the local search result quality and the off-site search result quality, and the local search result quality and the off-site search result quality are both the number of corresponding search results, the historical click rate, and the prediction. It may be expressed by at least one such as a click rate. For example, if the user's off-site search preference is greater than the third preference threshold and less than or equal to the second preference threshold, the off-site search result display area is placed in front of the local search result display area to retrieve information. The overall result display interface can be more tailored to the user's needs, as well as the information search results, which have a large number of local search results and a high historical CTR, a small number of offsite search results and a historical CTR. Is low, for example, when the user's off-site search preference is greater than the first preference threshold and less than or equal to the third preference threshold, the local search result display area should be placed in front of the off-site search result display area. You can As a matter of course, the user's off-site search preference and the local and off-site search result quality can be combined to determine the display order of the local search result display area and the off-site search result display area by a weighting algorithm, The information retrieval result display interface obtained in this way better suits the user's needs and further improves the user experience.

Considering that the number of local search results and/or the number of off-site search results may be large in the search results, displaying them all to the user at the same time may cause inconvenience to the user's reading and confuse the user. There is a nature. Therefore, in one embodiment according to the present invention, the local search result display area is set to a foldable display area, and the foldable display area displays only a part of the local search results in the folded state and in the unfolded state. Can display more local search results,
A control capable of controlling expansion or folding of the foldable display area is set in the foldable area, and when the user clicks the control, the expansion/contraction state of the foldable display area can be switched.

Correspondingly, the arrangement order of the local search results in the local search result display area and the number of local search results displayed in the folded state can be determined according to a preset rule. For example, in one embodiment according to the present invention, after merging the plurality of local search results into the local search result display area, further, in each of the local search result display area based on the predicted click rate of each local search result. The method may include determining the order of the local search results and determining the number of local search results displayed in the folded state of the local search result display area.

Specifically, you can calculate the predicted click-through rate for each local search result, and then sort the local search results based on the predicted click-through rate, with a specified number of top-sorted local searches. The result can be selected as the information displayed in the folded state.

Here, the predictive click rate (CTR: Click Through Rate) may be calculated by a click predictive algorithm, and an appropriate predictive model and a characteristic factor affecting the user's click are selected, and the historical click log data is used as a training set. Sort positions by combining, training model parameters and making click rate predictions for one new project.

In another embodiment according to the present invention, the off-site search result display area may be set to a foldable display area, and only a part of the off-site search result is displayed in the foldable display area. In the expanded state, more offsite search results can be displayed, and a control that can control expansion or folding of the collapsible display area is set in the collapsible area, and when the user clicks the control, The expansion/contraction state of the foldable display area can be switched.

Correspondingly, the arrangement order of the off-site search results in the off-site search result display area and the number of off-site search results displayed in the folded state can be determined according to a preset rule. For example, in one embodiment according to the present invention, after the plurality of off-site search results are merged into the off-site search result display area, the off-site search results are further calculated based on a predicted click rate of each of the off-site search results. Determining the order of each of the off-site search results in the display area and determining the number of off-site search results displayed in the collapsed state of the off-site search result display area.

Specifically, you can calculate the predicted clickthrough rate for each offsite search result, and then sort the offsite search results based on the predicted clickthrough rate, with a specified number of top-sorted multiples. Off-site search results can be selected as the information displayed in the collapsed state.

Considering that the display effect of the information search result is more suitable to the user's needs, the user experience is improved, and when specifically implemented, the user's off-site search preference, search context feature information, and local and off-site Based on the search result quality, it is possible to adaptively determine the arrangement order and the display number of the local and offsite results. For example, in one specific embodiment of the present invention, assuming that it is decided to preferentially display the local search result display area based on the user's off-site search preference and local and off-site search result quality, The number of results displayed in the folded state in the local search result display area may be estimated by the following method.

First, it is possible to set three constraint conditions, that is, a predicted click at each position in the local search result, a predicted click rate at the first position (the highest predicted click rate) in the off-site search result, and a smoothing parameter α. ..

When the local search result display area is preferentially displayed based on the above constraint conditions, the predicted click rate at the end of the local search results is α times larger than the predicted click of the first result of the off-site search results. The number of locations may be the number of local search result displays, i.e. the number of local search results displayed.

Similarly, when the off-site search result display area is preferentially displayed, at the position at the end of the off-site search results and the predicted click rate is α times larger than the predicted click of the first result among the local search results. The number may be the number of offsite search result displays, ie, the number of displayed offsite search results.

According to the above embodiment, it is possible to flexibly determine the number of search results displayed in the display area in the folded state, further satisfy the user needs, further optimize the display effect of the information search result page, and improve the user experience. ..

Specifically, the information search results are first sent to the client so that the client displays the local search results and the offsite search results in order after receiving the information search results. The search results and the off-site search results are sorted, the numbers to be displayed are determined, and the sort order of the information search results (that is, the display order of the clients with respect to the information search results) is as follows.

(1) If the user identifies that he/she has an off-site search intention and preferentially displays the local search results, the first preset number of the local search results in descending order of the predicted click rate is used. The local search results are preferentially displayed, and then a second preset number of offsite search results are displayed in descending order of predicted clicks of each offsite search result. Here, the difference between the predicted click rate of the last one displayed local search result and the predicted click rate α of the displayed first off-site search result is greater than 0, and the search process is mainly performed by the local search service. In consideration of the provision of the second number, the display number threshold value corresponding to the second preset number can be preset, and thereby the maximum number of off-site search results that can be displayed can be controlled.

(2) When the user has the intention of off-site search and preferentially displays the off-site search results, a third preset value is set in descending order of the predicted click rate of each off-site search result. The number of off-site search results is preferentially displayed, and then the fourth preset number of local search results is displayed in descending order of predicted clicks of each local search result. Here, the difference between the predicted click rate of the last one off-site search result displayed and the predicted click rate α times of the first displayed local search result is greater than 0, and the search process mainly uses the local search service. In consideration of the provision of the above, it is possible to preset the display number threshold value corresponding to the third preset number, thereby controlling the maximum number of off-site search results that can be displayed.

Considering that the information retrieval result obtained according to the embodiment of the present invention cannot meet the actual needs of the user 100%, the information retrieval result is transmitted to the client in one embodiment according to the present invention. After that, a user's click operation log for the information search result can be acquired, and thereafter, the off-site search preference degree determination model can be repeatedly optimized based on the click operation log. Here, the iterative optimization extracts search context feature information based on the click operation log, then adds it to a training sample of the off-site search preference determination model, and uses the new training sample to When iteratively trained using the off-site search preference model and iteratively optimize based on a large number of click operation logs, a cumulative effect is formed, whereby the off-site search preference model is later calculated. It may be to optimize the accuracy of the off-site search preference output during use. The continuous optimization continuously improves the accuracy of the identification with respect to the user's off-site search intention in the first embodiment of the present invention, and the display form of the information search result (for example, the local result display area after the optimization). And the display order of the off-site result display area may occur) can be further adapted to user needs.

Corresponding to the method shown in FIG. 1, as shown in FIG. 2, the information search device installed in the server according to the first embodiment of the present invention is
A search term receiving module 101 for receiving a search term transmitted from a user via a client,
A search intention identifying module 102 for identifying the search intention of the user based on the search word characteristic information and/or the search context characteristic information;
It is used to perform a search based on the search term using a search policy corresponding to the search intention and obtain an information search result related to the search term, and the information search result is a local search result and/or OFF. A policy search module 103 including site search results,
A search result transmission module 104 for transmitting the information search result to the client and causing the client to display the information search result.

In one embodiment according to the present invention, the search intention identification module 102 specifically acquires the feature information of the search term and/or the search context feature information, and the feature information of the search term and/or the search context feature information. Based on the search context feature information, to determine the off-site search preference level of the user by a pre-trained off-site search preference level determination model, and based on the off-site search preference level and preset preference level threshold conditions , Used to determine the search intention of the user.

In one embodiment according to the present invention, the off-site search preference level determination model includes a feature variable of a search term and/or a search context feature variable as an independent variable, an off-site search preference feature as a dependent variable, and an off-site search preference. Obtained by being trained by constructing a degree-determining model, obtaining training samples generated based on a plurality of search click history records, and training the off-site search preference degree model with the training supplement. ..

In one embodiment according to the invention, the off-site search preference model is constructed using a logical regression model.

In one embodiment according to the present invention, the off-site search preference determination model is specifically constructed by the following Equation 1.

In one embodiment according to the present invention, the information retrieval device further matches the search term with preset local and offsite entity word libraries to determine whether the search term has an offsite search potential. If there is a possibility of off-site search, an entity word matching module for performing a step of identifying the user's search intention based on the search word characteristic information and/or the search context characteristic information. Can be equipped.

In one embodiment according to the present invention, the policy search module 103 specifically, based on the search term, when the search intention is that the user has both local search intention and off-site search intention. It is used to perform local search and off-site search and obtain information search results including multiple local search results and multiple off-site search results.

In one embodiment according to the invention, the information retrieval device further comprises
A result merge module for merging the plurality of local search results into a local search result display area and merging the plurality of offsite search results into an offsite search result display area;
A display order determination module for determining the display order of the local search result display area and the offsite search result display area based on the off-site search preference of the user and/or the quality of the off-site search result;
A search result generation module for generating the information search result including the local search result display area and the off-site search result display area based on the display order may be provided.

In one embodiment according to the present invention, the local search result display area is a foldable display area. In this case, the information search device further determines the order of the local search results in the local search result display area based on the predicted click rate of each of the local search results, and in the folded state of the local search result display area. A local order determination module may be provided for determining the number of local search results displayed.

In one embodiment according to the present invention, the off-site search result display area is a foldable display area. In this case, the information search device further determines the order of the off-site search results in the off-site search result display area based on the predicted click rate of each of the off-site search result, and the off-site search result display area. An off-site order confirmation module for determining the number of off-site search results displayed in the collapsed state can be provided.

In one embodiment according to the invention, the information retrieval device further comprises
An operation log acquisition module for acquiring a user's click operation log for the information search result,
An iterative optimization module for iteratively optimizing the off-site search preference determination model based on the click operation log.

The information retrieval device and the information retrieval method are based on the same inventive concept and have the same beneficial effect, and therefore the description thereof is omitted here.

Corresponding to the method in FIG. 1, as shown in FIG. 3, the computer according to the first embodiment of the present invention includes a memory 1000 and a processor 2000, and the memory 1000 stores a computer program executable by the processor 2000. When the computer is executed by the processor 2000, the above information retrieval method is realized.

Specifically, the memory 1000 and the processor 2000 may be a general-purpose memory and a processor, and there is no particular limitation here, and when the processor 2000 executes a computer program stored in the memory 1000, It is possible to realize an information search method, which solves the problem that the related-art off-site search operation is complicated and inefficient, and when the user has the intention of off-site search. , It is possible to obtain the off-site search result without switching the city where the user is currently located, thereby effectively increasing the search efficiency and effectively improving the user experience.

Corresponding to the method shown in FIG. 1, the computer-readable storage medium according to the first embodiment of the present invention stores a computer program, and when the computer program is executed by a processor, realizes the above information retrieval method.

Specifically, the storage medium may be a general-purpose storage medium, for example, a moving magnetic disk, a hard disk, etc., and when the computer program in the storage medium is executed, the above information search method can be realized. , Which solves the problem that the related-art operation for performing off-site search is complicated and inefficient, and further switches the city currently located when the user has the intention of off-site search. Off-site search results can be obtained without having to do so, which can effectively improve search efficiency and effectively improve the user experience.

As shown in FIG. 4, in the second embodiment of the present invention, an information retrieval method is provided, the execution body of the method is a client, and the method is the information retrieval method for the server according to the first embodiment. It is implemented in combination with. For the related contents, refer to the description of the first embodiment above, and the description is omitted here. The second embodiment of the present invention is merely an example. The information retrieval method according to the second embodiment of the present invention includes steps S201-S204 and is specifically as follows.
In step S201, the search term input by the user is monitored.
In step S202, the search word is transmitted to the server.
In step S203, the information search result fed back from the server is received, and the information search result indicates that the server identifies the search intention of the user based on the characteristic information of the search term and/or the search context characteristic information. The information retrieval result includes a local retrieval result and/or an off-site retrieval result, which is obtained by performing a retrieval using a retrieval policy corresponding to the identified retrieval intention.
In step S204, the information search result is displayed.

In one embodiment of the present invention, the information search result includes a local search result display area, an off-site search result display area, and a display order. Here, the local search result display area includes a plurality of local search results, and the off-site search result display area includes a plurality of off-site search results.

Displaying the information search result includes displaying the local search result display area and the off-site search result display area on the same screen in the display order.

The information search method according to the second embodiment of the present invention is carried out in combination with the information search method according to the first embodiment, belongs to the same inventive concept, and carried out in combination with the first embodiment, As a result, it is possible to obtain a search result that further satisfies the user's intention, and when the user has the offsite search intention, the offsite search result can be obtained without switching the city in which the user is currently located. Can effectively improve the search efficiency and effectively improve the user experience.

Corresponding to the method shown in FIG. 4, as shown in FIG. 5, the information search device installed in the client according to the second embodiment of the present invention is
An input monitoring module 201 for monitoring a search term input by a user,
A search term sending module 202 for sending the search term to a server,
Information search results obtained by performing a search using a search policy corresponding to the user's search intention identified based on the search word feature information and/or the search context feature information fed back from the server A search result receiving module 203, wherein the information search results include local search results and/or off-site search results,
A search result display module 204 for displaying the information search result.

In one embodiment of the present invention, the information search result includes a local search result display area, an off-site search result display area, and a display order. Here, the local search result display area includes a plurality of local search results, and the off-site search result display area includes a plurality of off-site search results.

The search result display module 204 is specifically used to display the local search result display area and the off-site search result display area on the same screen in the display order.

The information retrieval apparatus according to the second embodiment of the present invention and the information retrieval method according to the second embodiment of the present invention are based on the same inventive concept and have the same beneficial effect, and therefore the description thereof is omitted here.

Corresponding to the method in FIG. 4, as shown in FIG. 6, the computer apparatus according to the second embodiment of the present invention includes a memory 3000 and a processor 4000, and the memory 3000 stores a computer program executable by the processor 4000. When the computer is executed by the processor 4000, the above information retrieval method is realized.

Specifically, the memory 3000 and the processor 4000 may be a general-purpose memory and a processor, and are not specifically limited here, and when the processor 4000 executes a computer program stored in the memory 3000, It is possible to realize an information search method, which solves the problem that the related-art off-site search operation is complicated and inefficient, and when the user has the intention of off-site search. , It is possible to obtain the off-site search result without switching the city where the user is currently located, thereby effectively increasing the search efficiency and effectively improving the user experience.

Corresponding to the method shown in FIG. 4, the computer-readable storage medium according to the second embodiment of the present invention stores a computer program, and when the computer program is executed by a processor, realizes the above information retrieval method.

Specifically, the storage medium may be a general-purpose storage medium, for example, a moving magnetic disk, a hard disk, etc., and when the computer program in the storage medium is executed, the above information retrieval method can be executed. , Which solves the problem that the related-art operation for performing off-site search is complicated and inefficient, and further switches the city currently located when the user has the intention of off-site search. Off-site search results can be obtained without having to do so, which can effectively improve search efficiency and effectively improve the user experience.

As shown in FIG. 7, the information search system according to the third embodiment of the present invention includes a server 1 and a client 2, the server 1 includes the information search device described in the first embodiment, and the client 2 is The information retrieval apparatus described in the second embodiment is included.

An information search system according to an embodiment of the present invention is realized based on the information search device according to the first embodiment and the information search device according to the second embodiment, and an operation for performing an off-site search in related technology is performed. It solves the problem of complexity and inefficiency, and when the user has the intention of offsite search, he can get the offsite search result without switching the city where he is currently located. The efficiency can be effectively increased and the user experience can be effectively improved.

The information retrieval device according to the embodiment of the present invention may be specific hardware of the device or software or firmware installed in the device. The device according to the embodiment of the present invention has the same principle of realization and generated technical effect as the embodiment of the method, and for the sake of simplicity, the contents not mentioned in the embodiment part of the device will be described. Reference may be made to the corresponding content in the method embodiments. For the sake of convenience and brevity, those skilled in the art can refer to the corresponding processes in the method embodiments for the specific operation processes of the above-described system, apparatus and unit, and the description thereof will be omitted here. Can be understood clearly.

It should be understood that in the embodiments provided by the present invention, the disclosed apparatus and method may be implemented by other methods. The embodiments of the apparatus described above are merely exemplary, for example, the division of the units is only division of logical functions, and there may be other division modes when actually implemented. Also, for example, units or components may be combined or integrated into another system, or some features may be ignored or not implemented. Also, the shown or discussed mutual couplings or direct couplings or communication connections may be indirect couplings or communication connections via some communication interface, device or unit, and may be electrical, mechanical or otherwise. It may be.

The units described as separating members may or may not be physically separate, and the members labeled as units may be physical units or physical It may not be a unit, ie it may be located in one place, or it may be distributed over multiple network elements. Depending on the actual needs, some or all of them can be selected to achieve the objective of the solution of this embodiment.

In addition, each functional unit in each embodiment according to the present invention may be integrated into one processing unit, each unit may physically exist alone, and two or more units may form one unit. It may be integrated into one unit.

The functions may be implemented in the form of software functional units and stored in one computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention may be realized essentially or in a part contributing to the prior art or a part of the technical solution in the form of a software product. A storage that includes some instructions for causing a computing device (which may be a personal computer, a server, or a network device, etc.) to perform the steps of all or part of the methods described in various embodiments of the invention. It may be stored on the medium. The pre-storage medium includes various media such as a U disk, a mobile hard disk, a read-only memory (ROM: Read-Only Memory), a random access memory (RAM: Random Access Memory), a magnetic disk, or an optical disk, which can store the program code.

Note that like symbols and letters represent like items in the following accompanying drawings, and when an item is defined in one accompanying drawing, it need not be further defined and interpreted in a later drawing. Should. Further, it is understood that terms such as “first”, “second”, and “third” are for distinctive explanation only and are not for indicating or implying relative importance. Should not be.

Finally, it should be noted that the above examples are merely specific embodiments of the present invention, which are for explaining the technical solutions of the present invention and not for limiting them. The protection scope of the present invention is not limited to this. Although the present invention will be described in detail with reference to the above examples, any person skilled in the art can still modify the technical solutions described in the above examples within the technical scope disclosed in the present invention, or Modifications can be easily conceived, or equivalent substitutions can be made for the partial technical features therein, these modifications, changes or substitutions being the essence of the corresponding technical solution. Without departing from the spirit and scope of the technical solutions in the examples. All should be included within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (20)

Information retrieval method,
The server receives the search terms sent by the user through the client;
The server identifying the user's search intent based on the search term feature information and/or the search context feature information;
The server uses a search policy corresponding to the search intention and searches based on the search term to obtain information search results related to the search term, wherein the information search result is a local search result and/or Or include offsite search results, and
The server transmits the information search result to the client and causes the client to display the information search result. Identifying the user's search intent based on the search term feature information and/or search context feature information comprises:
The server obtaining characteristic information of the search term and/or search context characteristic information;
The server determines the off-site search preference of the user by a pre-trained off-site search preference determination model based on the search word feature information and/or the search context feature information;
The method according to claim 1, further comprising: the server determining a search intention of the user based on the off-site search preference and a preset preference threshold condition. The off-site search preference determination model is
The server constructs the off-site search preference degree determination model, with the search word feature and/or the search context feature as independent variables and the off-site search preference feature as a dependent variable
The server obtaining training samples generated based on a plurality of search click history records,
The method of claim 2, wherein the server is trained and obtained by training the off-site search preference model with the training supple. The method of claim 3, wherein the server builds the off-site search preference determination model using a logical regression model. The server constructs the off-site search preference degree determination model by the following Equation 1,
The method further comprises
The server matching the search terms with preset local and off-site entity word libraries to determine whether the search terms are potentially offsite searchable;
When there is a possibility of off-site search, the server includes identifying the search intention of the user based on the feature information of the search term and/or the search context feature information. The method described. Searching using the search policy corresponding to the search intent and based on the search term, and obtaining information search results related to the search term,
If the search intent is that the user has a local search intent and an off-site search intent at the same time, the server performs a local search and an off-site search based on the search term, and a plurality of local search results and a plurality of off-site searches. The method of claim 1, including obtaining the information search results including site search results. The method further comprises
The server merging the plurality of local search results into a local search result display area and merging the plurality of offsite search results into an offsite search result display area;
The server determines a display order of the local search result display area and the offsite search result display area based on a user's off-site search preference and/or quality of local and off-site search results;
The method of claim 7, wherein the server generates the information search result including the local search result display area and the off-site search result display area based on the display order. The method further comprises
The server determines the order of the local search results in the local search result display area and the number of the local search results displayed in the collapsed state of the local search result display area based on the predicted click rate of each of the local search results. 9. The method of claim 8 including determining. The method further comprises
The server displays the order of the off-site search results in the off-site search result display area based on the predicted click-through rate of each of the off-site search results, and the off state displayed in the folded state of the off-site search result display area. The method of claim 8 including determining a number of site search results. The method further comprises
The server acquiring a click operation log of the user for the information search result,
The server iteratively optimizes the off-site search preference determination model based on the click operation log. Information retrieval method,
The client monitors the search terms entered by the user,
The client sending the search term to a server;
Obtained by the client by using a search policy corresponding to the user's search intent identified based on the characteristic information of the search term and/or the search context characteristic information fed back from the server. Receiving information search results, wherein the information search results include local search results and/or offsite search results;
The client displays the information search result. Displaying the information search results
The client displays a local search result display area and an off-site search result display area in the information search result on the same screen in the display order of the information search result, wherein at least one local area is displayed in the local search result display area. 13. The method of claim 12, including search results, including including at least one offsite search result in the offsite search result display area. An information retrieval device for a server,
A search term receiving module for receiving a search term sent from a user via a client;
A search intention identifying module for identifying the search intention of the user based on the characteristic information of the search word and/or the search context characteristic information,
It is used to search based on the search word using the search policy corresponding to the search intent and to obtain information search results related to the search word, wherein the information search result is a local search result and And/or a policy search module that includes off-site search results,
An information retrieval device for the server, comprising: a retrieval result transmission module for transmitting the information retrieval result to the client and causing the client to display the information retrieval result. An information retrieval device for a client,
An input monitoring module for monitoring the search terms entered by the user,
A search term sending module for sending the search term to a server;
Information search results obtained by performing a search using a search policy corresponding to the user's search intention identified based on the search word feature information and/or the search context feature information fed back from the server A search result receiving module, wherein the information search results include a local search result and/or an off-site search result,
An information search device for the client, comprising: a search result display module for displaying the information search result. Information retrieval system,
A server comprising the device according to claim 14;
A client comprising the device according to claim 15, and the information search system. 8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable by the processor, the method according to any one of claims 1-7 when the processor executes the computer program. The computer device according to claim 1, characterized in that: A computer device comprising a memory, a processor and a computer program stored in the memory and executable by the processor, the method comprising: performing the method according to claim 12 or 13 when the processor executes the computer program. The computer device characterized by the above. A computer-readable storage medium storing a computer program for implementing the method according to any one of claims 1-11 when the computer program is executed by a processor. A computer readable storage medium storing a computer program for implementing the method of claim 12 or 13 when the computer program is executed by a processor.